Degenerations of ideal hyperbolic triangulations (Q1925755)

Let \(M\) be the interior of a compact, connected, orientable 3-manifold with non-empty boundary consisting of tori and \(\mathcal{T}\) a topological ideal triangulation. That is, \(M\) is represented as an identification space of a disjoint union of 3-simplices whose vertices are removed and the identification is obtained by gluing pairs of 2-simplices of those 3-simplices to each other. In the paper under review, the author defines the deformation variety \(\mathfrak{D}(\mathcal{T})\) as a complex affine variety defined by the parameter equations and the hyperbolic gluing equations. The two systems of equations ensure that all the 3-simplices in \(\mathcal{T}\) can be realized as hyperbolic ideal tetrahedra and that they can be glued together so that a possibly incomplete hyperbolic structure on \(M\) is defined. Thus, a point of \(\mathfrak{D}(\mathcal{T})\) (with positive imaginary part in every coordinate) corresponds to a possibly incomplete hyperbolic structure on \(M\) (Theorem 2.4 and 2.5). Moreover, the set of ideal points \(\mathfrak{D}_{\infty}(\mathcal{T})\) of \(\mathfrak{D}(\mathcal{T})\) is defined to be \textit{G. M. Bergman}'s logarithmic limit set [Trans. Am. Math. Soc. 157, 459--470 (1971; Zbl 0212.53001)] of \(\mathfrak{D}(\mathcal{T})\). An ideal point corresponds to degenerations of some ideal hyperbolic tetrahedra which are related to a singular codimension-one measured foliation on \(M\). Also \(\mathfrak{D}_{\infty}(\mathcal{T})\) is canonically identified with a compact subset of the projective admissible solution space \(PF(\mathcal{T})\) of spun-normal surface theory. Every \(N\in PF(\mathcal{T})\) defines a singular codimension-one measured foliation \(\mathcal{F}\) on \(M\) such that if \(N\) has rational coordinate ratios, then an associated spun-normal surface with \(N\) corresponds to a union of leaves of \(\mathcal{F}\). The leaf space \(\widetilde{M}/\widetilde{\mathcal{F}}\) of the lifted foliation \(\widetilde{\mathcal{F}}\) on the universal cover \(\widetilde{M}\) is shown to be an \(\mathbb{R}\)-tree on which \(\pi_{1}(M)\) acts by isometries. Theorem 1.1 states that for \(\xi\in\mathfrak{D}_{\infty}(\mathcal{T})\) the action on the leaf space is non-trivial if \(\nu_{N}(\gamma)\neq 0\) for some peripheral element \(\gamma\in\pi_{1}(M)\), where \(N=N(\xi)\in PF(\mathcal{T})\) is the canonical admissible solution corresponding to \(\xi\) and \(\nu_{N}\) is the homomorphism from the fundamental group of each ``cusp-torus'' to \(\mathbb{R}\) defined by the author [Enseign. Math. (2) 54, No. 3--4, 329--380 (2008; Zbl 1214.57022)] via induced triangulation on the cusp-torus and Q-moduli. It is also shown that if \(\xi\) has rational coordinate ratios and the associated surface \(S(\xi)\) is non-compact, then \(S(\xi)\) is non-trivial, i.e., it is essential or can be reduced to an essential surface by performing compressions and then possibly discarding some components. A sequence in \(\mathfrak{D}(\mathcal{T})\) converging to an ideal point (in the sense of Bergman's logarithmic limit) is associated with the sequence of characters in the character variety of \(M\). It is shown that if the sequence of characters converges to an ideal point of the character variety in the sense of \textit{J. W. Morgan} and \textit{P. B. Shalen} [Ann. Math. (2) 120, 401--476 (1984; Zbl 0583.57005)], then the action on the leaf space is non-trivial (Theorem 1.2). Also, it is shown that a certain unboundedness of the sequence of the characters implies that the non-triviality of the surface \(S(\xi)\) and the duality to an ideal point of a curve in the character variety of \(M\). The author also shows the calculation and the description explicitly in the case of the figure eight knot complement.